Our study was approved by the Ethics Committee of Central South University (Changsha, China). Glioblastoma tissues (n=25) and normal brain tissues (n=15) were obtained from Xiangya Hospital of Central South University from April 2013 to January 2014. Written informed consent was obtained from all patients. The patients with glioblastoma, including 11 females and 14 males who ranged in age from 29 to 71 years, with a mean of 53.6 years, did not receive radiation therapy or chemotherapy before surgical resection. The non-tumor brain tissues were obtained by partial resections of normal brain tissue to reduce the increased intracranial pressure, when treating for severe head injury. All tissue samples were immediately snap-frozen in liquid nitrogen and stored at −80°C until use.

Human glioblastoma cell lines A172 and U87 were purchased from the Chinese Academy of Sciences (Shanghai, China), and cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (both from Life Technologies), 100 IU/ml penicillin and 100 IU/ml streptomycin at 37°C in a humidified atmosphere with 5% CO₂.

Total RNA of tissues or cells was extracted using TRIzol reagent (Life Technologies). A reverse transcription kit (Life Technologies) was used to convert RNA into cDNA, according to the manufacturer's instructions. For miR detection, real-time PCR was conducted using an miRNA qPCR detection kit (GeneCopoeia, Rockville, MD, USA) on an ABI 7500 thermocycler. The U6 gene was used as an internal reference. The real-time PCR primers for miR-663 were: forward, 5′-GTGCGTGTCGTGGAGTCG-3′ and reverse, 5′-TTTAGGCGGGGCG-3′. The real-time PCR primers for U6 expression were: forward, 5′-GCTTCGGCAG CACATATACTAAAAT-3′ and reverse, 5′-CGCTTCACGAAT TTGCGTGTCAT-3′. The PCR reaction conditions were 95°C for 5 min, followed by 40 cycles of 95°C for 15 sec and 60°C for 30 sec. The relative expression was analyzed by the 2^−ΔΔCt method (22).

The MTT assay was conducted to examine cell proliferation. Briefly, A172 and U87 cells in each group were plated at a density of 10,000 cells/well in 96-well plates. After being cultured for 6, 12, 24 and 48 h, the cells were incubated with MTT at a final concentration of 0.5 mg/ml for 4 h at 37°C. After the removal of the medium, 150 mM dimethylsulfoxide (DMSO) solution was added to dissolve the formazan crystals. The absorbance was read at 570 nm using a multi-well scanning spectrophotometer reader.

The tissues or cells were lysed in radio-immunoprecipitation assay buffer (Life Technologies). The protein was quantified using a BCA protein assay kit (Life Technologies). A total of 50 µg of proteins were separated on 10% SDS-PAGE gel and then transferred to a polyvinylidene difluoride (PVDF) membrane (Life Technologies), which was then incubated with Tris-buffered saline with Tween-20 containing 5% milk (Mengniu, Beijing, China) at room temperature for 3 h. The PVDF membrane was then incubated with primary antibodies: anti-TGF-β1, anti-MMP2, anti-MMP9, anti-E-cadherin, anti-N-cadherin, or anti-GAPDH antibody at 4°C overnight, followed by secondary antibodies conjugated with horseradish peroxidase at room temperature for 1 h. The protein bands were visualized by the Amersham ECL system (RPN998; GE Healthcare, Germany) and scanned. Data were analyzed by densitometry using Image-Pro Plus software 6.0 (Media Cybernetics, Rockville, MD, USA) normalized to GAPDH expression.

TargetScan was used to analyze the putative targets of miR-663. The full length of the wild-type (WT) of TGF-β1 3′-UTR was cloned into the psiCHECK2 luciferase reporter vector. The QuikChange Site-Directed Mutagenesis kit was used to generate a mutant type (MUT) of TGF-β1 3′-UTR, which was also cloned into the psiCHECK2 luciferase reporter vector. Lipofectamine 2000 was used to co-transfect miR-663 mimic or scramble miR (miR-NC) with psiCHECK2 vector inserted with WT or MUT TGF-β1 into the A172 and U87 cells, respectively. After incubation in a humidified atmosphere containing 5% CO₂ at 37°C for 48 h, a luciferase reporter assay was performed using the Dual-Luciferase reporter assay system (Promega, Madison, WI, USA), according to the manufacturer's instructions. Renilla luciferase was used for normalization. Each assay was performed in triplicate.

A wound-healing assay was performed to evaluate the cell migratory capacity of glioblastoma cells in each group. In brief, cells were cultured to full confluency. Wounds of ~1 mm in width were created with a plastic scriber, and cells were washed and incubated in a serum-free medium. After wounding for 24 h, the cells were incubated in a medium including 10% FBS. After being cultured for 36 h, the cells were fixed and observed under a microscope.

A cell invasion assay was performed using Transwell chambers (BD Biosciences, USA), which were pre-coated with Matrigel. A cell suspension containing 5×10⁵ cells/ml was prepared in serum-free media, and 300 µl of the cell suspension was added into the upper chamber. Then, 500 µl of DMEM with 10% FBS was added into the lower chamber. Cells were incubated for 24 h. Then, we used a cotton-tipped swab to carefully wipe off the cells that did not migrate or invade through the pores. The filters were fixed in 90% alcohol and stained by crystal violet, and observed under an inverted microscope (Olympus, Japan). After that, 0.5 g/l MTT was added and incubated at 37°C for 4 h. Then, the medium containing MTT was removed, and 50 µl of DMSO was added to each well. Following incubation at 37°C for 10 min, the optical density at 570 nm was measured using the Bio-Tek ELx800™ absorbance microplate reader (Bio-Tek, USA). The relative cell invasive capacity was determined.

Results shown in the figures are expressed as mean ± SD. Analysis of data was performed using GraphPad Prism 6 software. Student's t-tests or one-way ANOVA were used to analyze the significance of differences among groups depending on the experimental conditions. Statistical significance was evaluated by p-values of <0.05.

To reveal the role of miR-663 in glioblastoma, real-time RT-PCR was firstly conducted to examine the miR-663 levels in 25 cases of glioblastoma tissues, and 15 cases of normal brain tissues were used as controls. As shown in Fig. 1A, the expression levels of miR-663 were frequently reduced in the glioblastoma tissues, when compared with those in the normal brain tissues (P<0.01). We further found that miR-663 levels were also decreased in the glioblastoma A172 and U87 cells when compared to the level in the normal brain cells (Fig. 1B) (P<0.01). Based on these findings, we suggest that downregulation of miR-663 may be involved in glioblastoma progression.

Glioblastoma A172 and U87 cells were further transfected with the miR-663 mimic or the scramble miR as a negative control. After transfection, we conducted real-time RT-PCR to examine the miR-663 levels in each group. As shown in Fig. 2A, the miR-663 levels were significantly increased in the A172 and U87 cells transfected with the miR-663 mimic, when compared with those in the control group (P<0.01). However, transfection with miR-NC showed no effect on the miR-663 levels in the A172 and U87 cells, compared to the control group, respectively (P>0.05). These data indicated that the transfection efficiency was satisfactory.

We then studied the exact roles of miR-663 in the regulation of the malignant phenotypes of glioblastoma cells. MTT assay was conducted to examine the cell proliferation. As shown in Fig. 2B, upregulation of miR-663 levels significantly inhibited A172 and U87 cell proliferation, compared to the control groups, respectively (P<0.01). These data indicated that miR-663 had an inhibitory effect on glioblastoma cell proliferation.

We then performed a wound-healing assay and Transwell assay to examine the cell migration and invasion capacities of the glioblastoma cells in each group, respectively. Our data showed that restoration of miR-663 levels markedly suppressed A172 and U87 cell migration and invasion, compared to the control group, respectively (Figs. 3 and 4) (P<0.01). Therefore, our data demonstrated that miR-663 also suppressed the migration and invasion of glioblastoma cells.

TargetScan was further used to analyze the putative targets of miR-663, and TGF-β1 was predicted to be a target gene of miR-663. As indicated in Fig. 5A, perfect base pairing was shown between the seed sequence of mature miR-663 and the 3′-UTR of TGF-β1 mRNA. We further conducted a luciferase reporter assay to clarify this target relationship. After sub-cloning the wild-type (WT) or mutant type (MUT) of TGF-β1 3′-UTR (Fig. 5B) into a luciferase reporter psiCHECK2 vector, respectively, we co-transfected the miR-663 mimic or miR-NC with the psiCHECK2 vector inserted with the WT or MUT TGF-β1 3′-UTR into A172 and U87 cells. As shown in Fig. 5C, co-transfection with the WT TGF-β1 plasmid and the miR-663 mimic resulted in a significant decrease in the luciferase activity, when compared to the control group (P<0.01). However, co-transfection with the MUT TGF-β1 plasmid and the miR-663 mimic showed no alteration in the luciferase activity compared to the control group (Fig. 5C) (P>0.05). These data indicate that miR-663 directly binds to the 3′-UTR of TGF-β1 mRNA.

As miRs generally inhibit the protein translation of their targets (23), we further determined the effects of miR-663 on the protein levels of TGF-β1 in glioblastoma A172 and U87 cells. We firstly transfected these two cell lines with the miR-663 mimic or inhibitor, and then determined the miR-663 levels in each group. As shown in Fig. 6A, transfection with the miR-663 inhibitor significantly decreased the miR-663 levels in the glioblastoma cells, compared to the control group (P<0.01). Western blot assay was further conducted to examine the protein levels of TGF-β1 in each group. As shown in Fig. 6B, overexpression of miR-663 led to a significant decrease in the protein expression of TGF-β1 in the A172 and U87 cells (P<0.01). On the contrary, inhibition of miR-663 resulted in a marked increase in TGF-β1 protein levels in the A172 and U87 cells (Fig. 6B) (P<0.01). Based on the above data, we suggest that miR-663 negatively regulates the protein expression of TGF-β1 via direct binding to the 3′-UTR of TGF-β1 mRNA in glioblastoma cells.

As TGF-β1 was identified as a direct target of miR-663, and TGF-β1 has been demonstrated to play an oncogenic role in glioblastoma (24), we further investigated whether the role of miR-663 in glioblastoma was through mediation of TGF-β1. PcDNA3.1-TGF-β1 plasmid was trans-fected into miR-663-overexpressing A172 and U87 cells to reverse the reduced TGF-β1 levels. After transfection, western blot analysis was conducted. As indicated in Fig. 7A, the protein level of TGF-β1 was higher in the miR-663+TGF-β1 group, when compared with that in the miR-663 group (P<0.01). We next evaluated the cell proliferation, migration and invasion in each group. As shown in Fig. 7B, cell proliferation was upregulated in the U87 and A172 cells in the miR-663+TGF-β1 group, when compared with that in the miR-663 group (P<0.01), suggesting that TGF-β1 is involved in the miR-663-mediated cell proliferation in glioblastoma cells. Similar data were revealed in the cell migration and invasion analyses. Upregulation of TGF-β1 reversed the suppressive effects of miR-663 overexpression on glioblastoma cell migration and invasion (Figs. 8 and 9) (P<0.01). Collectively, these findings suggest that TGF-β1 acts as a downstream effector of miR-663 in the regulation of the malignant phenotypes of glioblastoma in vitro.

MMP2 and MMP9 are two key factors associated with the invasive capacity of cancer cells (25). Accordingly, we further conducted western blot analysis to evaluate their expression levels in each cell group. As shown in Fig. 10A, overexpression of miR-663 led to a significant decrease in the MMP2 protein level compared to that in the control group (P<0.01), which was then reversed after overexpression of TGF-β1 (P<0.01). These data suggest that MMP2 may be involved in miR-663/TGF-β1-mediated glioblastoma metastasis. In addition, we found no change in the MMP9 level in each group (Fig. 10A) (P>0.05). After that, we examined the protein level of E-cadherin, a key inhibitor of EMT (26). As shown in Fig. 10B, the protein expression of E-cadherin was notably upregulated in the miR-663-over-expressing glioblastoma cells, when compared to the control group (P<0.01), which was also reversed after upregulation of TGF-β1 (P<0.01). However, N-cadherin showed no difference in the cell groups (Fig. 10B) (P>0.05). Therefore, E-cadherin is suggested to also be involved in the miR-663/TGF-β1-mediated glioblastoma cell metastasis.